alphabetic reading development - bangor university...alphabetic reading development 4 2006, ch. 2)....

PR

IFY

SG

OL

BA

NG

OR

/ B

AN

GO

R U

NIV

ER

SIT

Y

Growth of Word and Pseudoword Reading Efficiency in AlphabeticOrthographiesCaravolas, Marketa

Journal of Learning Disabilities

DOI:10.1177/0022219417718197

Published: 01/09/2018

Peer reviewed version

Cyswllt i'r cyhoeddiad / Link to publication

Dyfyniad o'r fersiwn a gyhoeddwyd / Citation for published version (APA):Caravolas, M. (2018). Growth of Word and Pseudoword Reading Efficiency in AlphabeticOrthographies: Impact of Consistency. Journal of Learning Disabilities, 51(5), 422-433.https://doi.org/10.1177/0022219417718197

Hawliau Cyffredinol / General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/orother copyright owners and it is a condition of accessing publications that users recognise and abide by the legalrequirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of privatestudy or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal ?

Take down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access tothe work immediately and investigate your claim.

23. Apr. 2021

https://doi.org/10.1177/0022219417718197

https://research.bangor.ac.uk/portal/en/researchoutputs/growth-of-word-and-pseudoword-reading-efficiency-in-alphabetic-orthographies(7d3da71a-5267-4835-8d25-a19472d3af4b).html

https://research.bangor.ac.uk/portal/en/researchers/marketa-caravolas(5059ebe4-11a0-42cd-98c7-9eb9a4d290e3).html



https://doi.org/10.1177/0022219417718197

The Growth of Word and Pseudoword Reading Efficiency in Alphabetic

Orthographies: Impact of Consistency

Markéta Caravolas, PhD

School of Psychology, Bangor University, Bangor, LL57 2AS, UK

Running Head: ALPHABETIC READING DEVELOPMENT

Pre-print: Journal of Learning Disabilities

Corresponding Author:

Markéta Caravolas

Adeilad Brigantia

Bangor University

Bangor, LL57 2AS

United Kingdom

Email: [email protected]

Tel: +44 1248388334

Acknowledgements. This study was supported by Grant PITN-215961 – ELDEL from the

Marie Curie, Seventh Framework Programme to the author. I am grateful to Arne Lervåg for

valued advice on issues of statistical modelling, to members of the ELDEL Network for their

contributions to this study: senior scientists M. Mikulajová, G. Seidlová Málková, and

research fellows P. Mousikou, N. Salas, M. Schöffelová, and M. Litavský.

ALPHABETIC READING DEVELOPMENT 1

Abstract

Word and pseudoword reading are related abilities that are fundamental to reading

development in alphabetic orthographies. They are respectively assumed to index children’s

orthographic representations of words as acquired through the underlying ‘self-teaching

mechanism’ of alphabetic pseudoword decoding. However, little is known about the

concurrent growth trajectories of these skills in the early grades among children learning

different alphabetic orthographies. In the present study, between- and within-group latent

growth models of word and pseudoword reading efficiency were tested on data spanning first

and second grade, from learners of the inconsistent English, and consistent Czech and Slovak

orthographies. Several language-general patterns emerged. First, as expected, significant

growth was observed for both skills in all languages. Second, growth was faster for word than

pseudoword reading efficiency, and accordingly, strong lexicality effects that increased over

time were obtained across languages. Language-specific patterns were also found. In line

with predictions about the costs to learning of lower consistency orthographies, readers of

English experienced relatively slower growth on both reading skills. However, their lag was

smaller, and evident only at the latter two time points for word reading. In contrast, on

pseudoword reading, the English group performed considerably less well than their Czech

and Slovak peers at every time point. Thus, weak decoding skills were the main contributor

to the larger lexicality effects of the English group. These findings are considered within the

frame of recent theorizing about the effect of orthographic consistency on decoding as a self-

teaching mechanism in alphabetic reading acquisition.


Introduction

Measures of word and pseudoword reading efficiency are two ubiquitous indicators of

fluent and accurate word-level reading ability in alphabetic orthographies. They have been

contrasted in many languages, in experimental and cross-sectional studies seeking to uncover

the contribution of each skill to typical and atypical reading development. Yet, despite their

widespread use in reading research, and, the strong assumptions about their utility as indices

of reading proficiency (fluent word reading) and risk of reading failure (deficient pseudoword

reading), several aspects of their development remain unexplored. In particular, longitudinal

studies are lacking that examine concurrently how each skill grows in relation to the other,

and, how the consistency of letter-sound mappings in the orthography being learned might

influence their respective growth processes. To shed light on these questions, the present

study used latent growth modelling to investigate the growth of word and pseudoword

reading efficiency from first to second grade, among learners of the inconsistent English

orthography and learners of the relatively consistent orthographies of Czech and Slovak.

Background

Word reading efficiency refers to fluent and accurate reading of words; it is usually

assessed under time pressure, and is an important contributor to reading comprehension skill

(e.g., Perfetti, 2007). In alphabetic orthographies, efficient word reading arises most reliably

from children’s primary ability to decode printed words, that is, the ability to associate letters

(graphemes) more or less sequentially with their corresponding sounds (phonemes), and to

blend the sounds into accurate word pronunciations. Share’s (1995) self-teaching hypothesis

emphasizes decoding as the learning mechanism that underlies early alphabetic reading, and

proposes that, as children successfully apply the phonological (i.e., grapho-phonemic)

recoding procedure to newly encountered words, they build up word-specific orthographic

representations. Repeated decodings of specific words incrementally refine and strengthen


their orthographic representations, and consequently facilitate efficient word recognition

(e.g., Cunningham, Perry, Stanovich, & Share, 2002; Ehri, 2005, 2015). Thus, the decoding

process is understood to be the ‘sine qua non’ driver of robust orthographic representations

and of efficient word reading skills (Share, 1995).

Pseudoword reading (also referred to as nonword reading) is a widely used measure of

decoding ability. Written pseudowords are word-like in their graphotactic and phonotactic

structures; however, having neither lexical identity nor meaning, they present as novel items

that can only be read by the phonological recoding process. Within the self-teaching

hypothesis framework, and other compatible theories of reading development (e.g., Ehri,

2005; Perfetti & Hart, 2002), pseudoword reading tasks are used to assess children’s

decoding and orthographic learning skills (e.g., de Jong & Messbauer, 2011; Share, 2004).

Moreover, they are widely used to diagnose the nonword reading deficit (i.e., a phonological

processing deficit) in dyslexia (e.g., Herrmann, Matyas, & Pratt, 2006; Rack, Snowling, &

Olson, 1992).

Comparisons of word versus pseudoword performance on tasks of reading aloud typically

show an advantage in favor of word reading accuracy and/or speed, the lexicality effect. In

development, the emergence of the word reading advantage is thought to indicate that word

spellings have been lexicalized to some extent, that is, stored in memory in connection to an

existing lexical representation (e.g., Ehri, 2015). While questions about the exact nature of

these representations and how they are accessed remain hotly debated (e.g., see Rayner &

Reichle, 2010), they are tangential to our primary focus on the rate and pattern of growth of

the lexicality effect itself.

The lexicality effect has been reported in a large number of studies with child and adult

readers, skilled and less skilled, in many languages with alphabetic orthographies (e.g.,

Adelman, Sabatos-DeVito, Marquis, & Estes 2014; Sprenger-Charolles, Colé, & Serniclaes,


2006, ch. 2). Such studies show that its strength can be modulated by a variety of factors,

some of which are inherent to the participant, and others to properties of the stimulus

materials. Important participant variables are age and reading ability (which, in turn, may be

influenced by various cognitive factors, most typically phonological and/or grapho-phonemic

processing efficiency) (e.g., Rack et al., 1992; Van den Broeck, Geudens, & van den Bos,

2010). The item-level variables are numerous, but prominent among them are word

frequency, item length, and orthographic consistency (i.e., the extent to which letters and

sounds of printed words have one-to-one mappings) (Adelman et al., 2014; Rahbari &

Sénéchal, 2010). In the present study, the attribute of central interest was orthographic

consistency.

Alphabetic orthographies vary in terms of their system-wide consistency, and available

estimates show the English orthography to be the least consistent, while others such as Czech,

Slovak, and Spanish are relatively highly consistent (e.g., Caravolas & Samara, 2015). Within

orthographies, a number of factors may produce grapho-phonemic mapping (i.e., letter-

sound) inconsistency, including the extent to which they encode morphology, the frequency

and recency of spelling reform in the language (orthographies that rarely undergo reform tend

to contain more ‘irregular’ and ‘exception’ spellings reflecting archaic word pronunciations),

and, the degree of homophony in the language (see Kessler & Treiman, 2015). Although

some types of inconsistency (e.g., morphological encoding) may eventually provide insights

and certain advantages for the skilled reader/writer (e.g., Seidenberg, 2011), for the typical

novice reader, grapho-phonemic inconsistency is likely to present an obstacle to learning,

regardless of its origins.

One well-documented effect of orthographic consistency is its impact on the rate of word-

level reading growth: readers of more consistent orthographies learn to read more quickly

than readers of less consistent orthographies (e.g., Caravolas, Lervåg, Defior, Seidlová Málková,


& Hulme, 2013; Seymour, Aro, & Erskine, 2003). However, the influence of orthographic

consistency on the development of the lexicality effect is less clear. While no developmental,

language-general theory has focused specifically on the growth of these dual reading

processes, two theories have bearing on this issue. The psycholinguistic grain size theory

(Ziegler & Goswami, 2005) proposes that in consistent orthographies, readers can derive

pronunciations reliably using small-grain letter-sound mapping strategies whether decoding

words or pseudowords. In inconsistent orthographies, small-grain size decoding strategies--

while preferred for pseudoword reading--, are deemed less efficient for word reading, and are

therefore assumed to be supplemented by larger grain size mapping strategies (e.g., syllables,

rimes) that are salient in the spoken and written language. Extrapolating from this hypothesis,

readers of consistent orthographies should benefit little from the lexical status of words—

resulting in smaller lexicality effects; in contrast, the preference for different mapping

processes in word and pseudoword reading among learners of inconsistent orthographies

should result in larger lexicality effects. Along similar lines, the recent decoding stagnation

hypothesis, proposed by Van den Broeck and Geudens (2012) in the context of their work on

the nonword reading deficit in dyslexia, predicts larger lexicality effects for readers of

inconsistent orthographies than consistent ones, but also, that pseudoword decoding should

undergo gradual deceleration in growth specifically in inconsistent orthographies like

English. This is because learners of the latter systems should increasingly draw on their

word-specific orthographic knowledge when reading words, and with decreasing encounters

of novel word/pseudoword items, they exercise their phonological recoding skills less

frequently. In relatively consistent orthographies, the de-emphasis of phonological recoding

in word reading should be less pronounced, resulting in relatively faster growth of

pseudoword reading (no stagnation), and a smaller lexicality effect.


In line with the above theoretical predictions, several cross-linguistic studies have

reported larger lexicality effects for readers of English than those of a variety of more

consistent orthographies, and this seemed to be driven by English readers’ disproportionately

worse performance on pseudoword decoding (e.g., Landerl, Wimmer, & Frith, 1997; Patel,

Snowling & de Jong, 2004; Seymour et al., 2003; Wimmer & Aro, 2003). However, it is

difficult to generalize from these studies because most have used cross-sectional designs and

combined participant- and stimulus-based factors, language groups, and research

methodologies in myriad different ways. Moreover, they typically calculated the lexicality

effect on the basis of raw score differences rather than standardized differences that take

sample size and individual variation into account. Raw score effects may be misleading

because English readers may show greater variation in both word and pseudoword reading

than readers of more consistent orthographies, in particular when the consistent-orthography

groups show ceiling effects in accuracy, as is often the case. In one of the few single-

language studies investigating the development of the lexicality effect, Rahbari and Sénéchal

(2010) contrasted first to fourth graders’ reading efficiency of transparent (consistent) and

opaque (inconsistent) words, and pseudowords in the relatively consistent Farsi orthography.

Taking the view that successful reading experience drives the lexicalization of printed words,

which in turn facilitates word recognition, these authors predicted that children would

demonstrate lexicality effects of increasing magnitude from second grade onward for words

with the most transparent mappings, because transparency should accelerate the

establishment of lexicalized representations. On the other hand, they predicted no lexicality

effects in the primary grades for words that had opaque letter-sound mappings because the

orthographic representations for these should be more slowly acquired. Notably, Rahbari and

Sénéchal’s predictions did not include assumptions about children’s strategic changes in

grain-size mapping as a function of letter-sound consistency. Both predictions were borne


out by the data, and their finding of no lexicality effect for opaque words until adulthood

seemingly contradicted the general assumption of larger lexicality effects in less consistent

orthographies. Despite such mixed findings, the empirical evidence to date suggests that

language-specific differences are likely to emerge in the development of the lexicality effect,

and that these may be driven to a large extent by differences in orthographic consistency.

A Study of Word and Pseudoword Reading Growth in Three Orthographies

The above empirical studies also highlight the fact that to better understand the growth of

word and pseudoword reading skills and the influence of orthographic consistency on their

growth, longitudinal cross-linguistic studies are needed. The present study thus contrasted

concurrent growth of word and pseudoword reading efficiency, among three

language/orthography groups, British-English, Czech and Slovak, over a twelve month

period, from the end of first grade to the end of second grade. The linguistic and

orthographic features of the two Slavic languages have been described in our earlier work

(Caravolas et al., 2012); estimates of average letter-sound consistency in child-directed

printed words are approximately .92 for both languages (Kessler & Caravolas, 2011), while it

is .72 for English. However, the Slovak orthography was codified and reformed later than

Czech, resulting in its slightly greater system-wide phonographemic consistency (Caravolas,

in press).

The phase from end-Grade1 to end-Grade 2 was of particular interest because this is when

learners of many alphabetic orthographies, including English, are expected to reach a

watershed in reading development beyond which most become accurate, fluent, and

independent readers of words in their language (e.g., Caravolas et al. 2013; Rahbari &

Sénéchal, 2010; Standards & Testing Agency, 2015; Wimmer & Aro, 2003). Our previous

study of the growth of silent word reading, which followed progress among English, Czech,

and Spanish children over six time points from Reception/Kindergarten to the end of Grade 2


(Caravolas et al., 2013), showed that, despite similar levels of ability at the start of the study,

the English group had faster early skill growth due to their head start in reading instruction in

Reception Year (details in Participant section). However, once the consistent-orthography

readers had begun formal schooling, their rate of growth increased, overtaking the English

group by the end of Grade 1 and remaining ahead until the end of second grade. Based on

these findings, and on the general assumptions about the facilitating effect of orthographic

consistency on reading growth, a similar pattern was expected in the present study such that,

by the end of first grade and over the course of second grade, Czech and Slovak children

should have attained higher levels of reading-aloud proficiency and a faster rate of growth on

words as well as pseudowords than English children. Regarding the development of the

lexicality effect, an increase in its magnitude was expected over time for all groups, and the

decoding stagnation hypothesis in English was investigated.

Method

Participants

A total of 462 children (165 English – 48% girls; 124 Czech – 52% girls; 173 Slovak

– 48% girls) took part in the present study. They were participants on a broader, longitudinal

study comparing literacy development in five languages from Kindergarten/Reception Year

to the end of Grade 2. Children who were deemed to be readers on the basis of extreme

outlier scores on the Word Reading Efficiency test at first assessment in mid-

Kindergarten/Reception year (that is, 15-16 months prior to the start of the present study)

were deleted from the original data set; this represented 0.005% of the sample in English,

0.01% in Czech, and 0.06% in Slovak. English participants were recruited from primary

schools in the North of England; the Czech pupils came from basic schools across Bohemia,

and the Slovak pupils from basic schools in the Bratislava region. Details of the broader

samples along with demographic and educational information are reported in Caravolas et al.

(2012, 2013). The present samples comprised those children who contributed reading scores


to at least two of the three time points of this study, that is end-Grade 1 (May/June), mid-

Grade 2 (November/December), and end-Grade 2 (May/June).

Participant details for age and general ability are provided in Table 1. The English

group was on average one year younger than their Czech and Slovak peers, due to national

differences in the age of school entry, but had started to receive formal literacy instruction in

Reception Year, one year ahead of the other groups (who were at that point in Kindergarten).

However, previous research (Caravolas et al., 2012, 2013), with these same groups,

confirmed that they were well matched from the onset in Kindergarten/Reception Year

(England) on basic cognitive abilities as well as on the precursors of literacy. Children were

all monolingual speakers of the language of their country, and all had received phonically-

based instruction (see Caravolas et al., 2012 for further details).

-- Insert Table 1 about here --

Measures

Parallel measures of all tests were created (unless already existing) across the three

languages (see details below). The reading tests were administered individually as part of a

larger assessment battery, and in the same order at each time point.

Background measures

General ability. At the start of the broader study, when all participants were in mid-

Kindergarten/mid-Reception Year, the Vocabulary and Block Design subtests of the

Wechsler Preschool and Primary Scale of Intelligence for Children, WPPSI-III (Wechsler,

2003) were administered. For Czech and Slovak, the English versions were adapted and

standardized based on extended kindergarten-aged samples.

Reading measures

Attributes of Test items across Languages. Time-limited, one-minute, reading aloud

efficiency measures comprising lists of word and pseudoword items were administered at


each time point. The tests were designed to be long enough to ensure that growth in each

skill could be tracked over time without the risk of reaching ceiling effects, and allowing

the detection of individual variation within groups. Such measures are not designed to

investigate the processes underlying children’s word recognition or decoding, however,

they are appropriate for studies of skill growth in young children.

Both tests were designed to be comparable across languages in terms of the

constructs being measured (i.e., efficiency in reading familiar words versus efficiency in

decoding novel letter strings) and in the structure, length, and key characteristics of the

stimuli, while also reflecting age-appropriate printed words/letter sequences in each group’s

language. Summary statistics of the stimulus attributes for words and pseudowords in each

language are presented in Table A1 of the Appendix. A series of ANOVAs confirmed that,

(1) in line with the greater prevalence of monosyllabic words in English, the English stimuli

contained on average fewer syllables than the Czech and Slovak stimuli, which did not

differ from each other (words: F(2, 418) = 14.65, p < .001, 2 = .90; pseudowords: F(2,

418) = 11.35, p < .001, 2 = .90); in real terms, the differences were small, on average of

0.3 syllable; (2) importantly, in view of assumptions about the dominance of letter-by-letter

reading among young readers, especially of consistent orthographies (e.g., Share, 2008;

Ziegler & Goswami, 2005), the tests did not differ across languages on the average number

of letters per item (words: F(2, 418) =0.181, p > .8; pseudowords: F(2, 418) = 0.28, p > .7);

(3) English stimuli were on average somewhat more frequent than the Czech and Slovak

items (see means in Table A1), which did not differ from each other (F(2, 418) = 1459.33,

p < .001, 2 = .99). In sum, the test item attributes were comparable, and where small

differences occurred, they favored the English readers.

Word Reading-aloud Efficiency Test. In each language, 140 (144 in English)

words were selected from corpora of children’s school reading materials (Kessler &


Caravolas, 2011; Zeno, Ivens, Millard, & Duvvuri, 1995). They were of relatively high

frequency (see table A1), and occurred in the grade 1 and 2 corpora of each language; thus,

they were deemed to be familiar to 5-to-7-year-old children. Each list began with words

consisting of single letters (e.g., a, I, in English), progressing to more complex

monosyllables and then to two- and three-syllable words.

During individual administration, children read aloud as quickly as possible for 60

seconds from the word list, which was printed in Arial 16 pt. bold font and arranged in

three columns on both sides of an A4 sheet. The number of words read correctly (including

self-corrections) was recorded, while incorrectly or non-fluently read (that is, syllable- or

letter-sized units were pronounced but not blended) and skipped items received a score of 0.

This measure was first administered when the children were in Reception/Kindergarten, and

was found to have excellent test-retest correlations in all languages (r = .98, for English; r =

.88, r = .97 for Czech and Slovak, all ps < .01). Table 1 reports the test-retest reliabilites at

adjacent time points covered in the present study, which are again excellent.

Pseudoword Reading-aloud Efficiency Test. The pseudoword test, constructed in

the same way as the word test, was first administered at Time 1 of the present study. Its

items were derived from the word items by exchange of at least the word-initial letter, and

additional letters where appropriate. Crucially, the syllable structure and number of letters

remained the same across word-pseudoword pairs; occasional variations arose when

digraphs replaced single letter graphemes and vice versa (e.g., ‚child‘ ‚jild‘). The

number of pseudowords read plausibly in 60 seconds, including self-corrections, was

recorded. In English, any possible pronunciation of the graphemes, regardless of context

was considered correct; for example, for the pseudoword ‘jild’ both the pronunciations

[ʤaild], [ʤɪld] were accepted. Such variations were not relevant in the Czech and Slovak

versions. Implausibly or non-fluently read or skipped items received a score of 0. The


reliabilities, estimated from correlations of scores obtained at adjacent time points (Time 1

with Time 2, Time 2 with Time 3), are reported for each language in Table 1; all were

excellent.

Results

The descriptive statistics and reliabilities for the background variables and reading

measures across all time points and language groups are reported in Table 1. The mean scaled

scores on WPPSI Vocabulary and Block Design confirm that all three groups were of average

ability. The reading efficiency measures were designed to assess fluency more so than

accuracy. Accordingly, the error rates were low and decreased over time, as detailed in Table

A2. Although the English group committed on average more errors than their Czech and

Slovak counterparts, the low overall rates precluded further comparative error analyses.

Correlations between all observed reading variables across time points were very high

in all languages, with similar ranges as follows: English r = .75 - .93; Czech r = .78 - .91 and

Slovak r = .72 - .90. All ensuing growth curve analyses were carried out with Mplus 5.2

(Muthèn & Muthèn, 1998-2011) using Maximum Likelihood estimation. Missing values were

handled with Full Information Maximum Likelihood estimation. Both tests produced roughly

normal distributions across time points and language groups. Growth in word and

pseudoword reading-aloud efficiency was examined in two sets of unconditional latent

growth models, one focusing on between-language comparisons, the second on within-

language comparisons. The sample sizes were relatively small, especially for Czech, and

consequently, sample fluctuations in the residuals were anticipated. Across all models,

Heywood cases (negative, nonsignificant residual variances) were fixed to zero, or to be

equal over time points. Mean differences between growth constructs (intercepts, slopes) were

tested using the Wald test, where appropriate. Differences involving non-linear estimated

slopes were explored on the basis of the estimated factor loadings, and Cohen’s d effect sizes.


Patterns of Growth in Reading-aloud Efficiency across Orthographies

To compare reading skills across groups, a three-group unconditional latent growth

model of word reading efficiency, and another of pseudoword reading efficiency were

constructed, each tracking growth over the three time points, at roughly six-monthly

intervals.

Word Reading Efficiency. The three-group model with an initial reading status factor

(intercept) at Time 1 and a linear growth factor (slope) was estimated with residuals at Time

1 and Time 3 fixed to be zero across groups. The time scores, in year units, were fixed at 0

(Time 1), .5 (Time 2) and 1 (Time 3) for the Czech and Slovak groups, while for the English

group, the Time 2 score was estimated ( = 0.40, p < .001), reflecting nonlinear growth. This

model fitted the data very well, χ2 (8, N = 462) = 15.743, p = .05, Comparative Fit Index

(CFI) =.994, Tucker-Lewis Index (TLI) =.994, Root Mean Square Error of Approximation

(RMSEA) = .079 (90% CI = .001-.137), Standardized Root Mean Residual (SRMR) =.055,

and was accepted as the baseline model to test mean between-group differences in the

intercept and slope. The Wald test of mean differences in intercept was not significant (W462

(2) =2.74, p = .25), indicating that English, Czech and Slovak children showed comparable

levels of word reading aloud efficiency at the end of first grade. The mean difference in

slope between the Czech and Slovak children was significant (W297 (1) =3.973, p = .046)

indicating slightly faster growth for the Slovak group. The estimated slope for the English

sample could not be similarly tested, however, the estimated factor loading ( = 0.40, p

< .001) indicated a slower rate of growth by Time 2 than the other two groups. Thus, while

the Czech and Slovak children were acquiring word reading skills at a steady rate, having

accomplished 50% of their total year’s growth by mid-Grade 2, the English children had a

slower start, attaining only 40% of their year’s growth by Time 2, but then accelerated in the

second half of the year. Moreover, the English group showed smaller reading efficiency


gains between adjacent time points (Times 1 and 2: Cohen’s d = 0.33; Times 2 and 3: d =

0.55) than did the Czech and Slovak groups (ds ranging 0.65 – 1.00). The final model, in

which the intercepts were fixed to be equal over groups, provided a good fit to the data, χ2

(10, N = 462) = 18.47, p = .05, CFI =.994, TLI =.994, RMSEA = .074 (90% CI = .008-.126),

SRMR = .08. The estimated growth curves of this final model (Figure 1a), and the

corresponding means of the estimated growth factors (top panel of Table 2) illustrate that,

despite similar levels of word reading efficiency at the end of first grade, the rate of growth

over the ensuing year varied between groups, with the English group showing slower growth

than the Czech and Slovak groups, who, in turn, differed albeit negligibly.

-- Insert Table 2 --

Pseudoword Reading Efficiency. A similar model was estimated for pseudoword

reading efficiency. The residual variances were fixed to be equal within groups over time due

to Heywood cases. The time scores were fixed to 0 (Time 1), .5 (Time 2) and 1 (Time 3) for

the English and Czech groups, while for Slovak the Time 2 score was estimated ( = 0.61, p

< .001), reflecting nonlinear growth. This model provided a very good fit to the data, χ2 (8, N

= 462) = 15.54, p = .05, CFI =.993, TLI =.992, RMSEA = .078 (90% CI = .004-.136), SRMR

=.042, and was accepted as the baseline model to test mean between-group differences. The

Wald tests indicated large differences between intercepts for English and Czech (W289 (1)

=109.61, p < .001), English and Slovak (W338 (1) =91.50, p < .001), and a small, but

significant difference between the Czech and Slovak (W297 (1) =6.37, p < .05) groups.

However, despite the slightly lower starting level of the Slovak than the Czech group, both

appeared to have attained similar levels by the end of Grade 2. To test whether the latter

difference was significant, the model was respecified such that the intercept (0) was moved to

Time 3, while Times 1 and 2 were fixed to -1 and -.5, respectively. The Wald test showed, as

expected, that the Czech and Slovak groups did not differ in their end-Grade 2 attainments


(W297 (1) =0.35, p = .55). The test of slopes between the English and Czech groups revealed a

small but significant difference (W289 (1) =4.98, p < .05), indicating a slower rate of growth in

English. In contrast, the factor loading for the estimated slope for Slovak ( = 0.61, p < .001),

indicated relatively faster growth between Times 1 and 2 than the other two groups ( =

0.50); and faster growth was corroborated by larger effect sizes in pseudoword reading gains

at adjacent time points (Times 1 and 2: d = 1.10; Times 2 and 3: d = 0.61) relative to the

English and Czech groups, whose effect sizes were generally smaller (d = 0.40 to d = 0.71).

The final model, where the non-significant differences in intercept at Time 3 (Czech, Slovak)

were fixed to be equal provided a very good fit to the data, χ2 (9, N = 462) = 15.89, p = .07,

CFI =.994, TLI =.994, RMSEA = .070 (90% CI = .000-.126), SRMR =.044. Figure 1b and

the corresponding growth factor means (bottom panel of Table 2) show that, the English

group underwent a slower rate of growth, and unlike for word reading, had lower attainments

throughout the study relative to the consistent orthography groups, who reached similar

decoding efficiency levels by the end of Grade 2.

--Insert Figure 1a and 1b --

Growth in Word Relative to Pseudoword Reading-aloud Efficiency within Orthographies

The multigroup analyses revealed indirectly that all groups had lower levels of

attainment in pseudoword (Figure 1b) than in word reading (Figure 1a). The next set of

within-group models compared the relative growth of these skills as well as the size of the

lexicality effect as a function of time and orthographic consistency. A two-process latent

growth model was constructed for each language, through a model fitting process similar to

the multigroup analyses as regards fitting the intercepts, slopes, and the residuals (e.g.,

Heywood cases); in addition, the residual covariances were estimated between the parallel

observed variables (e.g., Time 1 word reading with Time 1pseudoword reading). The

lexicality effects were calculated using Cohen’s d on the estimated performance means and


standard deviations at each time point. Due to space constraints, these models are

represented in Figures 2a, 2b, 2c, along with the final model fit indexes, and lexicality effects

(full details are available from the author).

--Insert Figures 2a, 2b, 2c, about here--

As the graphs and captions in Figure 2 show, the two process models produced

adequate to good fits for the data; however, none accommodated equality constraints because

intercepts and rates of growth were significantly different between tasks in every language.

The lexicality effect at Time 1 was largest for the English group; however, it was very large

also in Slovak and Czech (d .90), and what is more, the effect increased in magnitude over

time in all languages. This was in all cases due to slower growth in pseudoword than in word

reading efficiency (see also the growth construct (G) means reported in Table 2).

Discussion

The effects of system-wide orthographic consistency on the developmentally critical

skills of word and pseudoword reading were considered among typically developing learners

of the relatively inconsistent English orthography with learners of the relatively consistent

Czech and Slovak orthographies. Each group’s ability levels, respective rates of skill growth,

and the magnitudes of the lexicality effect were assessed using similarly constructed and

highly reliable measures, three times from the end of Grade 1 to the end of Grade 2.

On the basis of cross-linguistic, cross-sectional studies reporting on word reading

aloud (e.g., Seymour et al., 2003; Wimmer & Aro, 2003), as well as our own longitudinal

findings for silent word reading efficiency (Caravolas et al., 2013), English children were

expected to achieve a poorer reading level compared to their Czech and Slovak peers by the

end of Grade 1 (the start of this investigation). However, the latent growth model revealed

that at end-Grade 1, all groups read words aloud with similar levels of efficiency, at a rate of

just over 47 words per minute, regardless of orthographic consistency. The relatively good


word reading skill of the English group is not likely to reflect specific educational effects,

such as being trained on tests of speeded list reading in order to attain recently introduced

early-reading benchmarks; such policies were not in place during the running of this study.

The English group’s result does contrast with our earlier finding of a clear lag in silent word

reading at end-Grade 1 (Caravolas et al., 2013), and suggests that orthographic consistency

effects manifest earlier in silent reading than in reading aloud. This in turn raises important

questions about the extent to which one can generalize about development across these two

reading modes (cf. Share, 2008).

These considerations aside, parity of the groups on word reading efficiency at the start

of this study provided an excellent baseline for further examinations of word as well as

pseudoword reading growth. Indeed by mid-Grade 2, the English children were on a slower

word-reading trajectory than their Czech and Slovak counterparts; and despite the English

group’s relative acceleration of growth in the second half of Grade 2, the gap between them

and the other groups continued to increase. Yet, it is important to note that in actual

performance terms, the difference in reading efficiency at end-Grade 2 was relatively small,

with the English group reading approximately seven words-per-minute fewer than their

Slavic peers. The Czech group showed somewhat slower growth than their Slovak

counterparts, but again, the words-per-minute difference was marginal. Whether the small

lag of the Czech group reflected random sampling differences, given their relatively small

sample size, or more interestingly, whether it reflected the impeding effects of the subtly

lower system-wide consistency of the Czech orthography (Caravolas, in press) is not clear,

and this issue awaits further research. However, if the latter were true, this would suggest that

even subtle differences in consistency can manifest in measurable differences in the early

word reading growth of different language/orthography groups.


The present paper (to our knowledge) reports the first longitudinal investigation of the

influence of orthographic consistency on the latent growth pattern of pseudoword reading.

Thus, expectations about the growth trajectories for this skill were guided by previous cross-

sectional studies, which consistently indicated poorer pseudoword decoding skills for readers

of English than those of more consistent orthographies (e.g., Landerl et al., 1997; Patel et al.,

2004; Seymour et al., 2003; Wimmer & Aro, 2003). Accordingly, at a point in reading

development when all three groups were reading words with similar levels of efficiency, the

English children were reading approximately 15 fewer pseudowords per minute than Czech

and Slovak readers, and this gap increased to approximately 18 items per minute at the end of

Grade 2 (see bottom panel of Table 2). The Slovak group had a marginally lower initial level

of performance than the Czech group, but had caught up by the end of Grade 2, again

indicating a slightly faster growth rate.

The within-language models were carried out primarily to assess the lexicality effect:

its magnitude and developmental changes, as a function of orthographic consistency. As

demonstrated in Figures 2a – 2c, and in line with theoretical predictions (Van den Broeck &

Geudens, 2010) and indications from cross-linguistic, cross-sectional studies (e.g., Patel et

al., 2004; Wimmer & Aro, 2003), smaller lexicality effects were obtained in Slovak and

Czech than in English, but, the effect increased over time in all groups. The analyses thus did

not provide evidence of Van den Broeck and Geudens’ (2012) specific hypothesis that typical

readers of less consistent orthographies, in this case English, might undergo a process of

decoding stagnation (i.e., a deceleration of pseudoword reading growth), due to their

decreasing reliance on decoding strategies during word reading. The small indication of

deceleration in Slovak, the group with arguably the highest relative orthographic consistency,

was more likely indicative of a transient plateauing of skill growth than of deceleration. The

latter findings must remain tentative, however, because within this three-measurement-points


design, it was difficult to robustly test the strength of the components of the non-linear

trajectory. In addition, it is possible that the developmental phase under study predated long-

term growth changes in any of the languages. We return to these issues below.

The results of the present study indicate that the costs of orthographic inconsistency

on alphabetic reading development are greater for the acquisition of decoding skills than for

the growth of real word reading, as evidenced by the relatively low scores of English children

on the pseudoword task. However, when considered within the theoretical framework of the

self-teaching hypothesis, the cloud of orthographic inconsistency of English may, after all,

have a silver lining. That is, to the extent that pseudoword reading is the proxy measure for

the foundational skill that drives orthographic learning and efficient word reading (Share,

1995), the present results may indicate that English readers are better able to capitalize on

their less developed decoding skills to access and acquire orthographic representations of real

words than their consistent-orthography-learning counterparts. Despite their considerable lag

in pseudoword reading at the end of first grade, the English children had levels of word

reading efficiency equivalent to their Czech and Slovak peers; thereafter, their word reading

lag remained relatively small.

The processes that gave rise to these language-specific growth patterns were not

investigated in the present study, nor was the quality of the orthographic representations

underpinning children’s word reading performance. Nevertheless the results invite some

interesting interpretations. For instance, the greater word reading advantage of the English

group may reflect their tendency to use more flexible (large and small grain) grapho-

phonological mapping strategies when reading words than when reading pseudowords, as

may be inferred from the psycholinguistic grain size theory (Ziegler & Goswami, 2005).

While possible, task-specific differences in grapho-phonemic mapping strategies are unlikely

to account fully for the present results in the light of accruing evidence that, already in the


early stages of learning to read, English children’s imperfect decoding skills are

supplemented by their extant lexico-semantic knowledge (Duff & Hulme, 2012; Laing &

Hulme, 1999; Nation & Cocksey, 2009). Thus relatively good word reading aloud

performance may be achieved by partial grapho-phonemic decoding in combination with

lexical knowledge, and, may be underpinned by more or less well specified orthographic

representations (see also Ehri, 2005, 2015; Landerl et al. 1997).

It is equally important to note that very large lexicality effects were consistently

obtained in the Czech and Slovak groups, clearly showing that learners of consistent

orthographies also draw large benefits from real word reading experience. The word reading

advantage in consistent orthographies presumably reflects some form of lexicalization (e.g.,

Rahbari & Sénéchal, 2010). Interesting recent evidence from German studies suggests that

this effect may in fact reflect children’s ever more efficient serial decoding strategies (as

opposed to whole word recognition) at least to fourth grade (e.g., Gagl, Hawelka & Wimmer,

2015; Rau, Moeller, & Landerl, 2014). Moreover, it is equally plausible, in line with a

language-general connectionist division of labour account of reading (e.g., Seidenberg,

2011), that, over and above the benefits of grapho-phonemic consistency, the word reading

advantage in consistent orthographies reflects dynamic contributions of lexico-semantic

knowledge to children’s more or less well established orthographic representations, as occurs

in English, although this contribution may be less important for successful word recognition

(e.g., Goswami, Ziegler, Dalton, & Schneider, 2001). The validity of this hypothesis, and the

relative contribution of each type of knowledge to word reading over the course of

development in different alphabetic orthographies remains to be investigated in future

longitudinal cross-linguistic studies.

Finally, the very high correlations observed in the present study between word and

pseudoword reading skills across languages and time points, despite clear between-language


differences in skill levels and growth rates, is consistent with the view that a tight functional

relationship between decoding skills and word reading skills comprises the self-teaching

mechanism in alphabetic reading development (Share, 1995).

The present study set the stage for further cross-linguistic investigations of early

reading growth. As noted earlier, growth curve modelling of three time points has limitations

in that non-linear growth patterns cannot be easily interpreted in such models. Also, while

the three time points examined here captured development in the run up to the putative

change in emphasis from ‘learning-to-read’ to ‘reading-to-learn’, it precluded the tracking of

more skilled reading development, that could moreover shed light on the decoding stagnation

hypothesis, and the predictive interrelationships between word and pseudoword reading skills

over a longer time span. Issues such as these should be investigated in future direct cross-

linguistic studies that span a longer learning period and sample a wider range of the

orthographic consistency spectrum.

References

Adelman, J., Sabatos-DeVito, M., Marquis, S., J., & Estes, Z. (2014). Individual differences in

reading aloud: A mega-study, item effects, and some models. Cognitive Psychology, 68, 113-160.

DOI:10.1016/j.cogpsych.2013.11.001

Caravolas, M. (in press). Learning to read in Czech and Slovak. In L. Verhoeven & C. E. Perfetti

(Eds.), Learning to read across languages and writing systems. Cambridge, UK: Cambridge

University Press.

Caravolas, M., Lervåg, A., Defior, S., Málková, G. S., & Hulme, C. (2013). Different patterns, but

equivalent predictors, of growth in reading in consistent and inconsistent orthographies.

Psychological Science, 24, 1398-1407. DOI: 10.1177/0956797612473122.

http://dx.doi.org/10.1016/j.cogpsych.2013.11.001


Caravolas, M., Lervåg, A., Mousikou, P., Efrim, C., Litavský, M., Onochie-Quintanilla, E., … Hulme,

C. (2012). Common patterns of prediction of literacy development in different alphabetic

orthographies. Psychological Science, 23, 678-686. DOI: 10.1177/0956797611434536.

Caravolas, M., & Samara, A. (2015). Learning to read and spell words in different writing systems. In

A. Pollatsek & R. Treiman (Eds.), The Oxford handbook of reading (pp. 326-343) New York, NY:

Oxford University Press.

Cunningham, A. E., Perry, K. E., Stanovich, K. E., & Share, D. L. (2002). Orthographic learning

during reading: Examining the role of self-teaching. Journal of Experimental Child Psychology,

82, 185–199.

De Jong, P. F., & Messbauer, V. C. (2011). Orthographic context and the acquisition of orthographic

knowledge in normal and dyslexic readers. Dyslexia, 17, 107-122. DOI: 10.1002/dys.427

Ehri, L. (2005). Development of sight word reading: Phases and findings. In M. Snowling & C.

Hulme (Eds.), The science of reading: A handbook (pp. 135-154). Oxford, England: Blackwell.

Ehri, L. (2015). How children learn to read words. In A. Pollatsek & R. Treiman (Eds.), The Oxford

Handbook of Reading (pp. 293-310). New York, NY: Oxford University Press.

Goswami, U., Ziegler, J. C., Dalton, L., & Schneider, W. (2001). Pseudohomophone effects and

phonological recoding procedures in reading development in English and German. Journal of

Memory and Language, 45, 648–664.

Herrmann, J. A., Matyas, T., & Pratt, C. (2006). Meta-analysis of the nonword-reading deficit in

specific reading disorder. Dyslexia, 12, 195–221. http://dx.doi.org/10.1002/dys.324

Kessler, B., & Caravolas, M. (2011). Weslalex: West Slavic lexicon of child-directed printed words.

Retrieved from http://spell.psychology.wustl.edu/weslalex

Kessler, B., & Treiman, R. (2015). Writing systems: Their properties and implications for reading. In

A. Pollatsek & R. Treiman (Eds.), The Oxford Handbook of Reading (pp. 10–25). New York, NY:

Oxford University Press.

Landerl, K., Wimmer, H., & Frith, U. (1997). The impact of orthographic consistency on dyslexia: A

German-English comparison. Cognition, 63, 315-334. DOI: 10.1016/S0010-0277(97)00005-X.

Muthèn, L. K., & Muthèn, B. O. (2010). Mplus user’s guide (5th ed.). Los Angeles, CA: Author.

http://dx.doi.org/10.1002/dys.324

http://spell.psychology.wustl.edu/weslalex

http://pages.wustl.edu/files/pages/imce/readingandlanguagelab/kessler_treiman_2015.pdf


Patel, T. K., Snowling, M. J., & De Jong, P. F. (2004). Learning to read in Dutch and English: A

cross-linguistic comparison. Journal of Educational Psychology, 96, 785–797.

http://dx.doi.org/10.1037/0022-0663.96.4.785.

Perfetti, C. A. (2007). Reading ability: Lexical quality to comprehension. Scientific Studies of

Reading,11, 357-383.

Perfetti, C. A. (2010). Decoding, vocabulary, and comprehension: The golden triangle of reading

skill. In M. G. McKeown & L. Kucan (Eds.), Bringing reading research to life: Essays in honor of

Isabel Beck (pp. 291-303). New York: Guilford.

Perfetti, C.A., & Hart, L. (2002). The lexical quality hypothesis. In L. Verhoeven. C. Elbro, & P.

Reitsma (Eds.), Precursors of functional literacy (pp. 189-213). Amsterdam/Philadelphia: John

Benjamins.

Rahbari, N., & Sénéchal, M. (2010). Learning to read and spell in Persian: A cross-sectional study

from grades 1 to 4. Developmental Psychology, 46, 1514-1527. DOI: 10.1037/a0020377

Rack, J. P., Snowling, M. J., & Olson, R. K. (1992). The nonword-reading deficit in developmental

dyslexia: A review. Reading Research Quarterly, 27, 29–53. http://dx.doi.org/10.2307/747832.

Rayner, K., & Reichle, E. D. (2010). Models of the reading process. Wiley Interdisciplinary Reviews:

Cognitive Science, 1, 787–799. doi:10.1002/wcs.68.

Seidenberg, M. S. (2011). Reading in Different Writing Systems: One Architecture, Multiple

Solutions In P. McCardle., J. Ren., & O. Tzeng. (Eds.), Across languages: Orthography and the

Gene-Brain-Behavior Link. Paul Brooke Publishing.

Seymour, P. H. K., Aro, M., & Erskine, J. M. (2003). Foundation literacy acquisition in European

orthographies. British Journal of Psychology, 94, 143–174.

http://dx.doi.org/10.1348/000712603321661859.

Share, D. L. (1995). Phonological recoding and self-teaching: Sine qua non of reading acquisition.

Cognition, 55, 151–218. http://dx.doi.org/10.1016/0010-0277(94)00645-2.

Share, D. L. (2004). Orthographic learning at a glance: On the time course and developmental onset

of self-teaching. Journal of Experimental Child Psychology, 87, 267–298.

http://dx.doi.org/10.1016/j.jecp. 2004.01.001.

http://www.pitt.edu/~perfetti/PDF/Reading%20Ability%20%28SSR%29%20scanned.pdf

http://www.pitt.edu/~perfetti/PDF/Reading%20Ability%20%28SSR%29%20scanned.pdf

http://www.pitt.edu/~perfetti/PDF/Decoding%20Beck%20festschrift%20chapt.pdf



http://www.pitt.edu/~perfetti/PDF/Lexical%20quality%20hypothesis-%20Hart.pdf



http://dx.doi.org/10.2307/747832

http://dx.doi.org/10.1016/0010-0277(94)00645-2


Share, D. L. (2008). On the Anglocentrism of current reading research and practice: The perils of

overreliance on an "outlier orthography. Psychological Bulletin, 134, 584–615. DOI:

10.1037/0033-2909.134.4.584

Sprenger-Charolles, L., Colé, P., & Serniclaes, W. (2006). Reading Acquisition and Developmental

Dyslexia. Hove, UK: Psychology Press.

Standards &Testing Agency (2015). Interim teacher assessment framework at the end of key stage

1. retrieved from

https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/461547/Interim_tea

cher_assessment_frameworks_at_the_end_of_key_stage_1_PDFA.pdf.

Van den Broeck, W. & Geudens, A. (2012). Old and new ways to study characteristics of reading

disability: The case of the nonword-reading deficit. Cognitive Psychology, 65, 414-456.


Van den Broeck, W., Geudens, A., & van den Bos, K. P. (2010). The nonword reading deficit of

disabled readers: A developmental interpretation. Developmental Psychology, 46, 717–734.

http://dx.doi.org/10.1037/a0019038.

Wechsler, D. (2003). Wechsler Preschool and Primary Scale of Intelligence—Third UK Edition

(WPPSI-IIIUK). Oxford, England: Psychological Corp.

Wimmer, H., & Aro, M. (2003). Learning to read: English in comparison to six more regular

orthographies. Applied Psycholinguistics, 24, 621-635.

Zeno, S.M., Ivenz, S.H., Millard, R.T., & Duvvuri, R. (1995). Educator's word frequency guide.

Brewster, NY: Touchstone Applied Science Associates.

Ziegler, J. C., & Goswami, U. (2005). Reading acquisition, developmental dyslexia, and skilled

reading across languages: A psycholinguistic grain size theory. Psychological Bulletin, 131(1), 3–

29. http://dx.doi.org/10.1037/0033-2909.131.1.3.

https://www.gov.uk/government/uploads/system/uploads/attachment_



Table 1

Participant Details and Means, Standard Deviations and Reliabilities for all Variables at all

Time Points for all Groups

English Czech Slovak

M (sd) Reliability M (sd) Reliability M (sd) Reliability

Age (months) at T1

75.01 (3.69)

-- 87.28 (3.86) -- 86.80 (3.71) --

WPPSI Vocabulary

(Scaled score) 9.64 (2.64)

10.15 (3.14) -- 9.99 (2.95) --

WPPSI Block Design

(Scaled score) 10.21 (3.10)

10.38 (2.72) -- 10.28 (2.99) --

Word Reading

Efficiency Time 1 49.06 (26.31) .93

1 48.91 (16.13) .83

1 46.24 (14.85) .86

1

Word Reading


2 65.07 (18.66) .87

2 62.21 (17.45) .90

2

Word Reading

Efficiency Time 3 71.09 (23.56)

--- 77.86 (19.28) -- 77.91 (18.02)

--

Pseudoword Reading


3 36.55 (11.51) .83

3 33.86 (9.57) .78

3

Pseudoword Reading


4 46.12 (12.63) .82

4 45.09 (10.75) .88

4

Pseudoword Reading

Efficiency Time 3 34.62 (16.74)

-- 52.61 (14.37) -- 52.20 (12.67) --

Note. 1

= correlations between 1 Minute Reading at first and second time point; 2 =

correlations between 1 Minute Reading at second and third time point; 3 = correlations

between 1 Minute Pseudoword Reading at first and second time point ; 4 = correlations

between 1 Minute Pseudoword Reading at second and third time point.


Table 2

Means, Standard Deviations of the Unconditioned Multigroup Growth Models for Word and

Pseudoword Reading Aloud Efficiency.

Growth

Constructs

Word Reading Efficiency by Language

English Czech Slovak Differences in means

Mean (SD)

Initial Status (I)

Growth (G)

47.52 ** (26.51**)

22.41** (14.02**)

47.52 ** (16.19**)

28.98** (10.09**)

47.52 ** (14.91**)

31.57** (11.63**)

Eng = Cz = Sk

Eng < Cz < Sk

Pseudoword Reading Efficiency by Language

Mean (SD)

Initial Status (I)

Final Status (F)

Growth (G)

21.16** (13.35**)

34.63**(16.21**)

13.48** (6.91**)

37.04** (10.33**)

52.56** (13.18**)

15.88** (4.83)

33.87** (8.72**)

52.56**(11.99**)

18.51** (7.02**)

Eng < Cz, Sk, Cz > Sk

Eng < Cz = Sk

Eng < Cz < Sk

Note. Means in bold did not differ over groups and were hence fixed to be equal over groups.

Variance in mean constructs and their standard deviations: *p < .05; **p < .01.


Appendix 1

Table A1

Stimulus Attributes of Word and Pseudoword Reading Efficiency Tests as a Function of

Language Group.

Language

English

(na = 144)

Czech

(na = 140)

Slovak

(na= 140)

Words Mean (SD) Mean (SD) Mean (SD)

Group

Differences

Mean Syllables/Item 1.33 (0.50) 1.63 (0.51) 1.61 (0.52) Eng < Cz = Sk

Mean Letters/Item 4.31 (1.61) 4.30 (1.30) 4.21 (1.32) ns

Mean Frequency - SFIb 69.65 (6.42) 63.81 (7.89) 64.43 (7.68) Eng> Cz = Sk

Pseudowords

Mean Syllables/Item 1.35 (0.53) 1.61 (0.49) 1.61 (0.52) Eng < Cz = Sk

Mean Letters/Item 4.32 (1.60) 4.23 (1.26) 4.20 (1.32) ns

Note: a: n refers to the number of items in the test for each language.

b: SFI = Standardized

Frequency Index derived from Zeno et al, (1995) for English and from Kessler & Caravolas,

(2011) for Czech and Slovak.


Appendix 2

Error Analyses

The error data on the Word and Pseudoword Reading Aloud Efficiency Tests were first

checked for extreme outliers, and these were trimmed to 2.5 SD above each group’s mean at

each time point. The percentage of extreme outliers ranged from 0.6% to 4.4% across groups

for word reading, and from 0.6% to 4.1% for pseudoword reading.

Table A2.

Errors Means and Standard Deviations for Word and Pseudoword Reading Aloud Efficiency

Tests as a Function of Language Group and Time.

Word Errors Pseudoword Errors

T1 M (SD) T2 M (SD) T3 M (SD) T1 M (SD) T2 M (SD) T3 M (SD)

English 5.14 (4.71) 3.62 (3.63) 2.33 (2.88) 7.05 (6.12) 6.45 (5.42) 5.87 (5.74)

Czech 2.13 (2.33) 1.03 (1.58) 0.64 (1.03) 3.85 (3.43) 3.22 (3.41) 2.81 (3.18)

Slovak 2.24 (2.22) 1.33 (1.54) 0.96 (1.26) 3.27 (2.35) 3.62 (3.41) 2.67 (2.21)

All distributions significantly positively skewed (p < .001) after trimming.


Figure 1b

Estimated Means for the Unconditional 2-Factor Growth Model of Nonword Reading Aloud Efficiency

End-Grade 1 Mid-Grade 2 End-Grade 2

FIGURE 1a

Figure 1a

Estimated Means for the Unconditional 2-Factor Growth Model of Word Reading Aloud Efficiency

End-Grade 1 Mid-Grade 2 End-Grade 2

FIGURE 1b


Figure 2a, 2b, 2c

Estimated Means for the Unconditional 2-Process Growth Models of Word Reading-aloud and

Nonword Reading-aloud Efficiency in English (2a), Czech (2b), and Slovak (2c), with Cohen’s d

estimates of the lexicality effect at each time point, and model fit indexes. On the X axis, 0 = End-

Grade 1, 0.5 = Mid-Grade 2, and 1 = End-Grade 2.

d (T1) = 1.42 d (T2) = 1.45 d (T3) = 1.89

Figure 2a

Figure 2b

d (T1) = 0.90 d (T2) = 1.22 d (T3) = 1.50

d (T1) = 1.00 d (T2) = 1.23 d (T3) = 1.69

Figure 2c

χ2 (5, N = 165) = 12.16, p = .03, CFI =.994, TLI =.983, RMSEA = .093 (90% CI = .025-.161),

SRMR =.041.


SRMR =.034.


SRMR =.063.

alphabetic reading development - bangor university...alphabetic reading development 4 2006, ch. 2)....

Documents